Promoter methylation analysis of DKK2 may be a potential biomarker for early detection of cervical cancer

Abstract Background Dickkopf 2 (DKK2) plays an important role in multiple cancers. Its potential value in the clinical diagnosis of cervical cancer has remained unclear. Objectives To investigate the expression and promoter methylation levels of DKK2 in cervical cancer and their clinicopathological associations. Methods We used the Gene Expression Omnibus, Oncomine, Cancer Genome Atlas, and University of ALabama at Birmingham CANcer data analysis databases, reverse transcription-PCR, and methylation-specific PCR analysis to predict and examine the expression of DKK2 mRNA and DKK2 methylation levels in cell lines and cervical cancer tissues from 79 patients with cervical cancer and 63 with cervical precancerous lesions including 25 with low-grade squamous intraepithelial lesions (LSIL) and 38 patients with high-grade squamous intraepithelial lesions (HSIL). Results DKK2 mRNA expression was downregulated in all cancer cell lines and cervical cancer tissues, whereas hypermethylation of DKK2 was higher in cervical cancer tissue samples. DKK2 methylation in cervical cancer was significantly higher than that in HSIL (χ2 = 8.346, P = 0.004), whereas DKK2 methylation in HSIL was significantly higher than that in normal cervical samples (χ2 = 7.934, P = 0.005) and in LSIL samples (χ2 = 4.375, P = 0.037). DKK2 silencing caused by its promoter hypermethylation was confirmed by treatment with the methyltransferase inhibitor 5-Aza-dC in cell lines. Patients with lymph node metastasis exhibited increased promoter methylation frequency (χ2 = 5.239, P = 0.022) and low DKK2 mRNA expression (χ2 = 3.958, P = 0.047) compared with patients with no lymph node metastasis. Patients with high-risk human papillomavirus infection exhibited increased promoter methylation frequency (χ2 = 6.279, P = 0.015). Conclusions DKK2 epigenetic changes of DKK2 may play a key role in the development of cervical cancer, suggesting that DKK2 hypermethylation could be used as a triage test for screening, early diagnosis, or risk prediction of cervical cancer.

Cervical cancer is the most common gynecological malignancy and is especially prevalent in young women [1]. While cervical cancer is often curable if detected at an early stage, the overall survival in patients with advanced cervical cancer remains poor. Therefore, better treatments and molecular markers related to carcinogenesis and progression that can help clinical practice are needed urgently. Screening is key for the early detection of precancerous lesions and cancer. ThinPrep cytologic tests (TCT) and human papillomavirus (HPV) detection are 2 important and commonly used methods to screen for cervical cancer [2]. TCTs are inexpensive, but their performance depends on several subjective human factors, which can lead to misdiagnosis and missed diagnosis, and poor accuracy. The sensitivity for detecting HPV is high, but most cases of HPV infection are transient, leading to poor specificity. Cytological screening strategies combined with HPV testing have contributed to a substantial reduction in the incidence and mortality of cervical cancer [3]. However, overall, the positive predictive effects are moderate, as only a small subset of people with HPV will eventually develop cervical cancer. Gene methylation analysis is a nonmorphological molecular detection method, which can provide an objective and potential shunt method for high-risk HPV-positive women. Triage tests using DNA methylation may be helpful to differentiate women who are at high risk of rapidly developing cervical cancer from women with low risk.
DNA methylation of tumor-related genes is closely related to early tumor progression, in a cancer type specific manner, and may provide information that can aid early screening of cervical cancer [4]. Aberrant promoter methylation, and epigenetic changes in tumor suppressor genes, either inactivating or silencing, are associated with tumorigenesis and progression [5][6][7][8]. In humans, DNA methylation occurs almost exclusively at the carbon 5 position on cytosine residues in CpG dinucleotides, which are concentrated in distinct GC-rich regions called "CpG islands" (CGIs) [8]. CpG island hypermethylation of tumor suppressor genes has been linked to the development of numerous human cancers, including cervical cancer [9,10]. The Wnt/b-catenin signaling pathway is known to be regulated by several secreted antagonists, such as members of the Dickkopf (DKK) protein family (DKK1, DKK2, DKK3, and DKK4), and has a function in embryonic development and tumorigenesis [11]. DKK2, located at 4q25, functions to produce an antagonist of canonical Wnt/b-catenin through its receptor low-density lipoprotein-receptor-related protein 5/6 (LRP5/6). The DKK2 promoter has a typical CpG island and is, therefore, under epigenetic regulation through promoter CpG methylation. DKK1 and DKK2 epigenetic silencing through promoter methylation have been observed in multiple cancers, including colorectal cancer, breast cancer, Ewing sarcoma, and gastric cancer [12][13][14][15][16][17][18]. However, its precise cellular function in cancer remains elusive.
Although DKK2 plays a significant role in many tumors, its correlation with clinicopathological characteristics such as lymph node metastasis and HPV infection is still unknown in cervical cancer. Here, we evaluated the mRNA expression profiles of DKK2 and its epigenetic alterations in cervical cancer cell lines and in samples from patients with cervical cancer.

Data mining and analyses
To compare gene expression differences between healthy donors and patients with cervical cancer, all clinicopathological data related to DKK2 expression profiles (GSE6791, GSE7803, GSE9750, and GSE7410) were carefully selected from the Gene Expression Omnibus (GEO) database, The Cancer Genome Atlas (TCGA) database (https://www.cancer. gov/about-nci/organization/ccg/research/structural-genomics/ tcga), and Oncomine websites (https://www.oncomine.org/ resource/login.html). GraphPad Prism was used to obtain the scatter diagram. DKK2 methylation expression data was mined in University of ALabama at Birmingham CANcer (UALCAN) databases (http://ualcan.path.uab.edu/) [19,20]. This study was conducted following the Reporting Recommendations for Tumor Marker Prognostic Studies (REMARK) checklist [21]. We analyzed a series of cervical specimens with histological diagnoses collected between May 2016 and November 2017 at Liaocheng People's Hospital (Shandong, China). At the time of surgical resection or cervical biopsy, 79 invasive cervical cancer tissues and 63 cervical intraepithelial neoplasia (CIN) tissues were obtained. Of the CIN tissues, 25 were classified as low-grade CIN (low-grade squamous intraepithelial lesions, LSIL) and 38 as highgrade CIN (high-grade squamous intraepithelial lesions, HSIL), and the latter included both CIN2 and CIN3. Normal cervical tissue (29 cases) was retrieved from patients with uterine leiomyomas who underwent a hysterectomy. All samples were stored at -80 °C until analysis. Genotyping was used to detect HPV16 and HPV18 viral DNA [20]. All cases had a confirmed diagnosis, as well as confirmed HPV typing and histopathological type. Exclusion criteria included samples from patients undergoing chemotherapy or radiotherapy, immunocompromised patients, pregnancy, diagnosis of other cancers, uterine cervix operations, and chronic or acute viral infection. Documented written informed consent was provided by all patients for tissue sample collection. The present study was approved by the medical ethics committee of Liaocheng People's Hospital, China

Reverse transcription-polymerase chain reaction
DKK2 mRNA expression was examined by reverse transcription-polymerase chain reaction (RT-PCR) analysis as previously described [18]. Total RNA was extracted from cell lines, CIN, normal cervical tissues, and tumor samples using TRIzol reagent (Invitrogen). DKK2 expression level was calculated relative to that of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) in each sample by RT-PCR. Primer pairs used for mRNA expression and the size of PCR products are described in Table 1. The products were separated in 2% agarose gels using a DL1000 marker set (catalog No. 3591A; Takara) marker set.

DNA extraction and bisulfite modification
DNA samples were extracted from tissues and cell lines. Genomic DNA (1 mg) was bisulfite-modified and purified following the guidelines described by the manufacturer. After desulfonation with NaOH and precipitation with ethanol, the final products were dissolved in 20 mL of Tris-EDTA buffer and stored at -80 °C.

Methylation-specific PCR
DKK2 gene methylation was determined using a Methylation-Gold Kit (Zymo Research). Primer pairs used for methylation analysis and PCR size products are shown in Table 1, as previously described [12,23]. Briefly, 5 min predenaturation occurred at 95 °C, 30 s denaturation for 35 cycles, 30 s annealing at 54 °C (unmethylated-specific PCR) or 58 °C methylation-specific PCR (MSP), and 10 min extension at 72 °C. As a positive control for methylation, DNA obtained from normal tissue was methylated in vitro with M.Sssl CpG methyltransferase (New England BioLabs). A water blank (no template DNA) was also included as a negative PCR control. The products were separated in 2% agarose gels using a DL1000 marker set (catalog No. 3591A; Takara) marker set.

Statistical analysis
SPSS Statistics for Windows (version 17.0) was used for all statistical analyses. Student t tests were performed for data mining, and χ 2 or Fisher exact tests were performed for RT-PCR and MSP analyses. P < 0.05 was considered significant.

DKK2 mRNA expression in cervical cancer cell lines and tumor specimens
We examined DKK2 mRNA expression in 79 normal cervical tissues (50 by data mining, 29 by RT-PCR), 9 cervical cancer lines (HT-3, C4-I, CaSki, MS751, C33A, SiHa, SW756, ME-180, and HeLa cell lines by data mining, HeLa, HT-3, C33A, and CaSki cell lines by RT-PCR), and 498 cervical cancer samples (419 by data mining, 79 by RT-PCR). Based on data from the GEO database (Figures 1A-D), DKK2 mRNA expression was lower in cervical cancer tissues compared with normal tissues (P < 0.0001 for Figures 1A, B, and D, P = 0.0004 for Figure 1C). However, based on data from the TCGA database (Figure 1E), the difference was not significant (P = 0.848), although the levels of DKK2 mRNA expression showed a similar trend to that of the GEO database. Compared with normal cervical epithelial cells, expression in cervical cancer cells including HT-3, C4-I, CaSki, MS751, C33A, SiHa, SW756, ME-180, and HeLa cervical cancer cell lines was lower, as shown in Figure 1F (P = 0.023).

Correlation of DKK2 inactivation with its promoter hypermethylation
We then examined the status of DKK2 promoter methylation in HeLa, HT-3, C33A, and CaSki cell lines by MSP. DKK2 hypermethylation was detected in all 4 cell lines (Figure 2A). Of these 4 cell lines, the C33A cell line showed the weakest DKK2 mRNA expression, although both unmethylated and methylated bands were detected. Correlation analysis between DKK2 silencing and promoter hypermethylation was performed by treating cell lines with 10 mM 5-Aza-dC induction for 4 d, followed by promoter methylation level analysis. DNAdemethylating agent treatment significantly led to DKK2 demethylation and re-expression of the transcript among all 4 cell lines analyzed (Figure 2B). We then examined DKK2 methylation status in cervical tissues. Methylation levels were significantly higher in cervical cancer tissues, as shown in TCGA samples (Figure 2C, P < 0.0001). MSP analysis was then performed to confirm the overall frequency of DKK2 methylation in cervical cancer. DKK2 promoter hypermethylation was detected in 41 of 79 cases (51.9%) of cervical cancer tissues. However, methylation status was only found in 10 of 63 CIN cases (12.7%), and no methylation status was detected in normal cervical cases ( Figure 2D). The positive rate of DKK2 methylation in cervical cancer was significantly higher than that in HSIL (χ 2 = 8.346, P = 0.004), whereas the positive rate of DKK2 methylation in HSIL was significantly higher than that in normal cervical samples (χ 2 = 7.934, P = 0.005) and in LSIL samples (χ 2 = 4.375, P = 0.037) ( Table 2). Positive DKK2 mRNA expression in cervical cancer cases without methylation (47.4%) was significantly higher than that of cervical cancer tissues with promoter hypermethylation (17.1%) (χ 2 = 8.368, P = 0.002).

Discussion
In the present study, we investigated DKK2 expression profiles and epigenetic alterations in cervical cancer. DKK2 mRNA expression was reduced in cell lines of cervical cancer and cancer tissues, whereas DKK2 hypermethylation was upregulated. In addition, DKK2 mRNA expression was restored after 5-Aza-dC treatment of cell lines of cervical cancer. The study suggests DKK2 silencing is strongly associated with its promoter hypermethylation.
The activated Wnt/b-catenin signaling pathway plays an important role in cervical cancer. CpG island promoter hypermethylation has been shown to inactivate extracellular Wnt antagonists in cervical cancer [22,23]. CpG island methylation can be widely found in the human genome, but only a subset of loci plays important roles in tumorigenesis. These implicated genes are known to be involved in multiple cellular signaling pathways such as apoptosis, cell cycle regulation, development, differentiation, invasion, and metastasis. As CpG island methylation occurs early and methylated alleles can be detected in a sensitive manner in carcinogenesis, detection of methylation may be a promising tool for early detection of cancer [24]. DKK2, a Wnt antagonist, contributes to tumorigenesis in multiple cancers [12,13,[25][26][27][28]. In the present study, RT-PCR showed reduced expression of DKK2 mRNA in cervical cancer compared with either normal cervical samples or HSIL samples. Further, through MSP analysis DKK2 is found to be methylated in most patients diagnosed with cervical cancer, while it is not in HSIL and normal cervical samples. Collectively, the data demonstrate that DKK2 is predominantly methylated in cervical cancer. DKK2 Methylation of the DKK1 promoter in patients with cervical squamous cell carcinoma is related to high-risk HPV infection and histological differentiation, tumor size, lymph node metastasis, and International Federation of Gynecology and Obstetrics (FIGO) staging, while the degree of methylation of DKK1 is not related to the type of high-risk HPV infection [14]. Paired PAX1 methylation was found to be a valuable biomarker for cervical cancer screening, a commonly used method in our hospital, with a 77% sensitivity and 92% specificity of CIN3+ versus normal [29]. A high methylation rate of DKK2 was significantly associated with poor overall survival, and a multivariate Cox proportional hazards model revealed that methylation of DKK2 is an independent adverse prognostic factor [28]. In the present study, significant DKK2 hypermethylation was detected in lymph node-positive cervical cancer. The methylation rate of the DKK2 promoter was 76.5% in cervical cancer specimens with lymph node metastasis, and only 45.2% in cervical cancer specimens without lymph node metastasis, indicating a significant difference between the 2 groups. HPV infection accounts for over 90% of cervical cancer cases, and the high-risk types of HPV are associated with 87%-88% of squamous cell carcinomas [30]. A triage test using DNA methylation may be helpful to differentiate women who are at high risk of developing cervical cancer rapidly from women with low risk. In the case of HSIL, the combination of HPV genotyping and methylation marker analysis can not only overcome the limitations of cytological examination but also increase diagnosis accuracy. The present study indicated that among HR-HPV-positive cervical cancer patients, the methylation rate of DKK2 in HPV16/18-positive patients was more greatly enhanced than that in HPV16/18-negative patients.
We acknowledge that the present study also has some limitations. In the precancerous lesion group, the difference was not statistically significant. The correlation between DKK2 methylation level and HPV subtype needs to be investigated further by including a larger sample size. In addition, the population included in the present study was drawn from patients in hospitals, not from a population that would be screened. The real-world screening value of DKK2 methylation detection for cervical cancer needs further population-based research and discussion.

Conclusion
DKK2 epigenetic changes may play a key role in the development of cervical cancer, suggesting a potential value of DKK2 hypermethylation as a triage test for screening, early diagnosis, or predicting the risk of cervical cancer.
Author contributions. XZ and YZ contributed to the conception and design of the study. AL, JW, YW, and XM acquired the data, and XZ, YL, QC, and YZ analyzed it. XZ, AL, JW, and YW drafted the manuscript, and XM, YL, QC, and YZ critically revised it. All authors approved the final version submitted for publication and take responsibility for statements made in the published article.